Neurabins recruit protein phosphatase-1 and inhibitor-2 to the actin cytoskeleton.

Inhibitor-2 (I-2) bound protein phosphatase-1 (PP1) and several PP1-binding proteins from rat brain extracts, including the actin-binding proteins, neurabin I and neurabin II. Neurabins from rat brain lysates were sedimented by I-2 and its structural homologue, I-4. The central domain of both neurabins bound PP1 and I-2, and mutation of a conserved PP1-binding motif abolished neurabin binding to both proteins. Microcystin-LR, a PP1 inhibitor, also attenuated I-2 binding to neurabins. Immunoprecipitation of neurabin I established its association with PP1 and I-2 in HEK293T cells and suggested that PP1 mediated I-2 binding to neurabins. The C terminus of I-2, although not required for PP1 binding, facilitated PP1 recruitment by neurabins, which also targeted I-2 to polymerized F-actin. Mutations that attenuated PP1 binding to I-2 and neurabin I suggested distinct and overlapping sites for these two proteins on the PP1 catalytic subunit. Immunocytochemistry in epithelial cells and cultured hippocampal neurons showed that endogenous neurabin II and I-2 colocalized at actin-rich structures, consistent with the ability of neurabins to target the PP1.I-2 complex to actin cytoskeleton and regulate cell morphology.

The type 1 protein serine/threonine phosphatase or protein phosphatase-1 (PP1) 1 is highly conserved in all eukaryotes and regulates a variety of functions including protein synthesis, gene transcription, glycogen and lipid metabolism, muscle contractility, and learning and memory (1,2). These diverse functions are directed by the unique localization and substrate recognition by PP1 conferred via its association with regulatory or targeting subunits. Hormonal control of cellular PP1 activity is also mediated by covalent modification of targeting subunits and the endogenous protein inhibitors (3)(4)(5). With the discovery of a highly conserved PP1-binding motif, KIXF, in both targeting subunits and inhibitors (6), it was postulated that PP1 accommodated only one of these regulators, and PP1 displacement from the targeting subunits was required for its effective inhibition by endogenous protein inhibitors (3). This hypothesis was supported by studies of the prototypic PP1 complex composed of PP1 bound to the glycogen-targeting subunit, G M , which showed a decreased sensitivity to inhibitor-1 (I-1) (7,8) and DARPP-32 (9), two PKA-activated PP1 inhibitors. Thus, PKA, which phosphorylated G M within an RVSF sequence, coordinated PP1 displacement with the activation of I-1 to suppress the dephosphorylation of glycogen-bound phosphoproteins and promote glycogenolysis in skeletal muscle (3).
However, recent studies (10 -13) have identified several PP1 inhibitors that bind and inhibit targeted forms of PP1. I-1, for example, binds a PP1-targeting subunit, GADD34 (protein product of growth arrest and DNA damage-inducible gene 34), that regulates protein translation. As both regulators require a KIXF motif to bind PP1, the coexistence of the two proteins on a single PP1 catalytic subunit occurs through both direct interactions between the regulators and their independent association with the PP1 catalytic subunit (10).
These findings prompted us to evaluate cellular PP1 complexes for the presence of multiple regulators. We focused on inhibitor-2 (I-2), which associated with several PP1-binding proteins from rat brain extracts. Two of these I-2 interacting proteins were identified as neurabin I and neurabin II/spinophilin, neuronal actin-binding proteins that target PP1 to F-actin and regulate cell morphology (14). Both in vitro and in vivo studies suggested that PP1 bridged the association of I-2 with neurabins. Moreover, neurabin enhanced the targeting of PP1 and I-2 to polymerized F-actin. Finally, immunocytochemistry confirmed that I-2 and neurabin II (15,16) colocalized at actin-rich structures in cultured epithelial cells and primary hippocampal neurons. Together, the studies identified a multiprotein complex containing neurabin, PP1, and I-2 that potentially regulates dephosphorylation events at the actin cytoskeleton.
For Western immunoblotting, proteins were subjected to electro-phoresis on 10% polyacrylamide gels in the presence of SDS-PAGE. The gels transferred electrophoretically to polyvinylidene difluoride membranes, which were blocked in TBS (10 mM Tris-HCl, pH 7.5, 150 mM NaCl) plus 0.05% Tween and 4% (w/v) milk. The membranes were then incubated with primary antibody at 4°C overnight, washed with TBS plus 0.05% Tween, and incubated with appropriate secondary antibodies for 1 h at room temperature. The immunoreactive proteins were visualized with chemiluminescence (PerkinElmer Life Sciences).
The F460A mutation was introduced in rat neurabin I using the QuikChange Site-directed Mutagenesis Kit (Stratagene), the oligonucleotide primer, 5ЈGAAATCCCAGCAAATAGGAAAATTAAGGCTA-GTTGTGCTCCG3Ј, and its complementary reverse primer. The pEGFP-C1 (Clontech) plasmid encoding GFP-NrbI-(1-552) was used as template. The BglII and BamHI fragment of the mutant NrbI cDNA was then inserted into a plasmid encoding GFP-NrbI-(1-1095) to introduce the F460A mutation into the full-length NrbI.
Preparation of Brain Lysates-Rat brain (Pel-Freeze) was homogenized using a Dounce homogenizer in 50 mM Tris-HCl, pH 7.5, containing 5 mM EDTA, 5 mM EGTA, 10 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, and 1 g/ml each of aprotinin, leupeptin, and pepstatin. Following centrifugation at 100,000 ϫ g for 60 min, the lysate was rapidly frozen in liquid N 2 and stored at Ϫ80°C. Deoxycholate extracts were prepared in a similar manner using the same buffer containing 1% w/v deoxycholate (DOX). The DOX lysate was dialyzed against 100 volumes of 50 mM Tris-HCl, pH 7.5, containing 1 mM EDTA, 1 mM EGTA, 10 mM NaCl, 0.1 mM phenylmethylsulfonyl fluoride, and 0.1 mM benzamidine for 3 h with one change of buffer prior to storage at Ϫ80°C. PP1 Far Westerns-PP1-binding proteins were identified using the previously described Far Western procedure (24). Briefly, recombinant rabbit PP1␣ was covalently modified using the digoxigenin labeling kit (Roche Molecular Biochemicals). Samples containing PP1-binding proteins were separated on 10% (w/v) SDS-PAGE and electrophoretically transferred to a polyvinylidene difluoride membrane. The membrane was blocked with 4% (w/v) milk dissolved in 10 mM Tris-HCl, pH 7.5, and 150 mM NaCl, and the PP1-binding proteins were detected by incubating the membrane with digoxigenin-conjugated PP1␣ followed by an anti-digoxigenin antibody (Roche Molecular Biochemicals). Antibody binding was visualized by the alkaline phosphatase reaction.
Pulldown Assays-For sedimentation or pulldown assays with GST fusion proteins, glutathione-Sepharose (25-l bed volume) was equilibrated with TBS (10 mM Tris-HCl, pH 7.5, containing 150 mM NaCl). GST proteins were added, and the volume was adjusted to 300 l with TBS prior to incubation for 1 h at 4°C. The beads were washed 2 times with TBS, and rat brain lysate (4 mg of total protein) was added, and the mixture was incubated for 1 h at 4°C. The beads were washed 4 times with NETN-250 (10 mM Tris-HCl, pH 7.5, 1 mM EDTA, 250 mM NaCl, and 0.5% Nonidet P-40), and the bound proteins were eluted using 25 l of SDS sample buffer prior to SDS-PAGE on 8.75, 10, or 12% (w/v) acrylamide gels. Pulldowns using hexahistidine-tagged proteins were undertaken similarly using Ni ϩ /NTA-agarose (Qiagen), and the protein-bound beads were washed with PBS (10 mM sodium phosphate pH 7.5 containing 150 mM NaCl) containing 10 mM imidazole prior to elution with SDS sample buffer. In indicated pulldowns, rat brain lysates were preincubated with 10 M microcystin-LR (MC; Alexis Biochemicals) for 30 min prior to the addition of GST fusion proteins. The beads were washed 4 times (5-10 min each) with NETN-250 containing 5 M MC prior to elution of bound proteins with SDS sample buffer.
To isolate endogenous PP1 complexes, MC-Sepharose (25-l bed volume containing 1 mg of MC per ml of Sepharose) was incubated with rat brain DOX lysate (4 mg of total protein) for 1 h at 4°C. The beads were washed 4 times with NETN-250, and bound proteins were eluted with SDS sample buffer.
GST-G M -(1-240) phosphorylated in vitro with PKA was also used in pulldowns as described previously (25). Briefly, GST-G M (5 g of total protein) was immobilized on glutathione-Sepharose and phosphorylated with PKA catalytic subunit (10 units) for 30 min at 30°C (22). The reaction was stopped by washing beads with PBS containing 1 mM dithiothreitol, and beads with 2 g of bound protein were incubated with NIH3T3 cell lysates (22). After extensive washing, bound proteins were eluted with SDS sample buffer and analyzed by SDS-PAGE.
For cosedimentation of purified proteins, GST-NrbII-(354 -494) (5 g of total protein) was incubated with glutathione-Sepharose for 1 h at 4°C. The beads were washed twice with TBS, and increasing concentrations of purified skeletal muscle PP1 (26) dialyzed into 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 15 mM ␤Ϫmercaptoethanol, and bovine serum albumin (1 mg/ml) or the buffer alone were added and mixtures incubated for 1 h. Following this, purified rabbit skeletal muscle I-2 (1 g of total protein) was added and the incubation continued for 1 h. The beads were then washed 3 times with TBS, and the proteins were eluted with 25 l of SDS sample buffer and analyzed by SDS-PAGE.
For quantitative assays of ␤-galactosidase expression, diploid strains were grown in selective liquid media to A 0.6. An aliquot (1.5 ml) of the cultures was subjected to centrifugation, and yeasts were resuspended in 200 l of Z buffer (100 mM H 3 PO 4 , 10 mM KCl, 1 mM MgSO 4 , pH 7.0). The cells were permeabilized in 50 l of 0.1% (w/v) SDS and 50 l of chloroform by vortexing three times for 30 s each, and the lysates were assayed for ␤-galactosidase using O-nitrophenyl S-D-galactoside as described (31).
Immunoprecipitation-HEK293T cells, grown to near-confluency in 6-well plates, were transfected with 1.5 g of DNA encoding HA-hI-2-(1-197), 1.5 g of GFP-NrbI-(1-1095) DNA, and 6 l of LipofectAMINE (Invitrogen), according to the manufacturer's instructions. After 24 h, cells were washed with PBS and lysed in 500 l of RIPA buffer (10 mM sodium phosphate pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 0.5% deoxycholate, and 0.1% SDS) for 15 min on ice. Cells were scraped and the lysates cleared by centrifugation at 20,000 ϫ g for 3 min. 1 l of Living Colors TM polyclonal anti-GFP antibody (Clontech) was added and the mixture incubated for 1 h (14). A 1:1 slurry of protein A-agarose (Bio-Rad) and protein G-Sepharose (Sigma) were added for 1 h, and beads were washed 4 times with NETN-250 prior to solubilizing the immunoprecipitates in SDS sample buffer and analysis on SDS-PAGE.
Actin Sedimentation-F-actin sedimentation was undertaken using the Actin Binding Protein Biochem Kit (Cytoskeleton) (14). Briefly, 40 g of G-actin was polymerized in 40 l of polymerization buffer. Combinations of 7.5 g of hexahistidine-NrbI-(1-516), 5 g of PP1, and 2 g of I-2, both purified from rabbit skeletal muscle, were added in a total volume of 60 l, and the mixture was incubated for 30 min at room temperature. The polymerized actin was sedimented by centrifugation at 150,000 ϫ g for 1.5 h at 24°C, and the pellet containing F-actin and bound proteins or the soluble fraction was analyzed by SDS-PAGE. To establish actin-mediated sedimentation, we included a depolymerization step that solubilized the F-actin-bound proteins in 200 l of 1 mM Tris-HCl, pH 7.5, 0.5 mM EDTA, 0.5 mM EGTA, and 0.1% ␤-mercaptoethanol, followed by dialysis against 1,000-fold excess volume of the same buffer for 3 h at room temperature and centrifugation at 150,000 ϫ g for 1.5 h. This procedure solubilized Ͼ90% of F-actin and the actin-associated proteins, which were analyzed by SDS-PAGE.
Immunocytochemistry-Madin-Darby canine kidney (MDCKII) cells were grown to confluence on fibronectin-coated coverslips in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen). Cells were washed with PBS containing 1 mM CaCl 2 and 0.5 mM MgCl 2 , permeabilized with a 1:1 mixture of methanol/acetone for 10 min at Ϫ20°C, rinsed twice with PBS containing 0.05% Tween 20 (PBS-T), and incubated in a 10% fetal bovine serum/PBS-T blocking solution for 1 h at room temperature. Sheep polyclonal antibody against human I-2 (17) and rabbit polyclonal antirat spinophilin/NrbII (Upstate Biotechnology, Inc.) were diluted in PBS-T at 1:500 and 1:400 respectively, and applied to the coverslips for 1 h at room temperature. Cells were rinsed 3 times with PBS-T (5 min each) before staining with secondary antibodies (fluorescein isothiocyanate-labeled donkey anti-sheep and rhodamine-labeled goat anti-rabbit) diluted in PBS-T. Coverslips were rinsed with PBS-T and PBS prior to mounting on glass slides with 10 l of Vectashield mounting medium (Vector Laboratories).
Primary rat E19 embryo hippocampal neurons were plated at a density of 2 ϫ 10 5 cells/60-mm dish on polylysine-treated coverslips and cocultured with rat glial cells for 14 days (32) in minimum essential medium (Invitrogen) supplemented with 10% horse serum. The cultures were rinsed with PBS, fixed in 4% paraformaldehyde for 15 min at room temperature, and quenched in 50 mM ammonium chloride for 10 min at room temperature prior to permeabilization in PBS containing 0.2% Triton X-100 for 2 min. The coverslips were rinsed with PBS, blocked with 3% bovine serum albumin for 1 h at room temperature, and stained as described above using an Alexa Fluor 488-labeled donkey anti-sheep antibody (Molecular Probes).
Digital images were captured using a Nikon Microphot-SA epifluorescence microscope with Nikon Plan Apo 60x/1.4 oil immersion objective and appropriate filters for fluorescein isothiocyanate, rhodamine, and DAPI and a Hamamatsu Orca C4742-95 digital camera. The image files created using Openlab software (Improvision) were contrast-enhanced, pseudocolored, and merged using Photoshop 6.0 software.
Structural Determinants in Neurabins Required for I-2 Binding-To define domains in rat NrbI and NrbII that dictated their association with I-2, we conducted pulldowns from rat brain DOX lysates using several recombinant neurabins. GST and GST-hI-2-(1-205) (20 g of total protein) were immobilized on glutathione-Sepharose and used in pulldowns from rat brain extracts (4 mg of total protein) as described under "Materials and Methods." After extensive washing, the bound proteins were eluted and analyzed with a sample of the initial lysate (input) on 10% SDS-PAGE. A shows far Westerns or overlays of pulldowns from cytosol and deoxycholate extracts from rat brain using digoxigenin-PP1 as a probe. B, pulldowns from DOX extracts were analyzed by Western immunoblotting using antibodies against PP1, NrbI (neurabin I), NrbII (neurabin II/spinophilin), GADD34, NF-L (neurofilament light chain), and PSD-95 (postsynaptic density protein of 95 kDa). and GST-GADD34-(513-674), a fragment of human GADD34 (10, 36), bound PP1 from rat brain DOX lysates. However, only GST-G M -(1-240) also recruited I-2 (Fig. 3A). This demonstrated that a subset of PP1-targeting subunits in mammalian tissues associate with both PP1 and I-2. PKA phosphorylation of G M at serine 67 within the PP1-binding motif abrogated PP1 binding both in vivo and in vitro (37), and in pulldowns from NIH3T3 cell extracts using GST-G M -(1-240) phosphorylated in vitro by PKA, neither PP1 nor I-2 binding was observed (Fig.  3B). This suggested that, as noted with neurabins, I-2 associates with G M via PP1, and phosphorylation within the PP1docking motif regulates the association of a PP1⅐I-2 complex with G M .
Microcystin-LR, a PP1 Inhibitor, Attenuates I-2 Binding to Neurabin/PP1-MC, a hepatotoxin, binds the PP1 catalytic site and inhibits enzyme activity. The high affinity of MC for PP1 and other serine/threonine phosphatases has prompted the use of MC-Sepharose for affinity isolation of numerous cellular phosphatase complexes (38). MC-Sepharose concentrated PP1 associated with NrbI and NrbII from rat brain lysates, but these complexes contained no detectable I-2 (Fig.  4A). This was consistent with earlier observations suggesting a competition between MC and I-2 for PP1 binding (20). Thus, MC-Sepharose excluded PP1 complexes in which the PP1 catalytic site was occupied by I-2.
To investigate this further, we also undertook pulldowns with GST-neurabins from DOX lysates in the presence and absence of 10 M MC (Fig. 4B). The addition of MC prevented binding of both PP1 and neurabins to GST-hI-2-(1-205). In the converse experiment, GST-NrbI-(374 -516) and GST-NrbII-(354 -494) bound both PP1 and I-2 from DOX lysates. Although addition of 10 M MC abolished I-2 binding, both GST-neurabins showed an increased association with PP1 under these conditions (Fig. 4B). These in vitro studies not only suggested that I-2 regulated levels of PP1 recruited by neurabins but also supported the notion that I-2 bound the active site of the PP1-neurabin complex and generated an inactive phosphatase complex.

FIG. 2. Structural basis for I-2 binding to neurabins.
A, His-hI-2 polypeptides (20 g of total protein) were immobilized on Ni ϩ /NTAagarose and incubated with rat brain DOX lysates. Ni ϩ /NTA-agarose alone was used as control. B, GST-hI-4 proteins (20 g of total protein) were immobilized on glutathione-Sepharose and used in pulldowns from DOX lysates. The bound proteins and a sample of the Input were subjected to SDS-PAGE and immunoblotted with antibodies against PP1, NrbI, and NrbII.

FIG. 3. Structural determinants in neurabins for I-2 binding.
A, GST fusion proteins (20 g of total protein) were immobilized on glutathione-Sepharose and used in pulldowns from DOX lysates. The bound proteins and sample input were subjected to SDS-PAGE and immunoblotted using antibodies against PP1 and I-2. B, GST-G M -(1-240) (2 g of protein) immobilized on glutathione-Sepharose was incubated with ATP-Mg in the presence (ϩ) or absence (Ϫ) of PKA (as described under "Materials and Methods") prior to addition of NIH3T3 cell extracts. GST-G M -(1-240) was visualized by protein staining with Coomassie Blue, and PP1 and I-2 were analyzed by immunoblotting with appropriate monospecific antibodies. C, GST-NrbII-(354 -494) (5 g of total protein) immobilized on glutathione-Sepharose was incubated with PP1 (0.35, 1.75, and 8.75 g of protein) or control buffer prior to the addition of purified rabbit skeletal muscle I-2 (1 g of protein). Beads were washed and the bound proteins eluted with SDS sample buffer and subjected to SDS-PAGE followed by immunoblotting with an anti-I-2 antibody.
Interaction of PP1 Catalytic Subunit with Neurabin I and I-2-A single gene (GLC7) encodes the yeast PP1 catalytic subunit, which shares 87% overall sequence identity to human PP1. Mutational analyses of GLC7 have provided insights into the interactions of PP1 with several mammalian regulators (29,30,39). By using a yeast two-hybrid assay, we analyzed the association of rat NrbI and human I-2 with 39 yeast PP1 mutants (27,28). Surprisingly, NrbI proteins containing the N-terminal actinbinding domain failed to bind PP1, and a deletion that yielded NrbI-(103-1095) was required to promote PP1 binding. Whereas a majority of PP1 mutants bound both regulators effectively, a subset of the mutants showed deficits in their association to hI-2-(1-205) or NrbI-(103-1095) (Fig. 5A). PP1(K259A,R260A) and PP1(D229A,R233A) bound hI-2-(1-205) similar to WT PP1, but PP1(K110A,K112A) showed a diminished association with hI-2-(1-205). This mutation, however, had little or no effect on PP1 binding to NrbI-(103-1095). In contrast, PP1(K259A,R260A) and PP1(D229A,R233A) showed severely diminished binding to NrbI-(103-1095). Enzyme assays using partially purified WT and mutant PP1 catalytic subunits correlated the loss of twohybrid interaction with a reduced sensitivity of PP1 to inhibition by I-2 (data not shown). The amino acids that selectively attenuated the association of PP1 with I-2 and/or NrbI were mapped on the three-dimensional structure of the PP1 catalytic subunit (Fig. 5B), and the data suggested some differences in PP1 docking by the two regulators.

. Disruption of neurabin-I-2 association by microcystin-LR.
A, rat brain DOX lysates were incubated with MC-Sepharose beads as described under "Materials and Methods." After extensive washing, bound proteins were eluted using SDS sample buffer and with the input subjected to immunoblotting with antibodies against PP1, I-2, NrbI, and NrbII. B, various GST fusion proteins (5 g of protein) were immobilized on glutathione-Sepharose prior to incubation with brain DOX extracts in the presence (ϩ) or absence (Ϫ) of 10 M MC as indicated under "Materials and Methods." Following their elution, the bound proteins were analyzed by immunoblotting with the indicated antibodies.

FIG. 5. Mapping neurabin and I-2 binding to PP1 catalytic subunit.
A, top, a yeast two-hybrid assay was performed with WT human inhibitor-2 fused to the GAL4 DNA-binding domain as bait, and binding was analyzed with WT and mutant PP1 catalytic subunits fused to the GAL4 transcriptional activation domain (see "Materials and Methods"). Expression of the ␤-galactosidase reporter gene was analyzed as enzyme activity shown with standard errors. A, bottom, rat NrbI-(103-1095) fused to the GAL4 DNA-binding domain was used as bait in the yeast two-hybrid assay to analyze its interaction with the PP1 catalytic subunits as described above. B shows the three-dimensional structure of PP1 catalytic subunit with the residues required for NrbI binding (blue) and those recognized by I-2 (red). Also included are the proposed docking site for the IKGI sequence essential for potent PP1 inhibition by I-2 and amino acids (green) that accommodated both NrbI and I-2; green colored amino acids include the hydrophobic groove that binds the conserved KIXF sequences and the ␤12-␤13 loop that binds both PP1 inhibitors and targeting subunits (48). in HEK293T cells. Anti-GFP immunoprecipitates contained equivalent levels of GFP-NrbI-(1-1095) (the full-length protein) and GFP-NrbI-(1-552) (Fig. 6). HA-hI-2-(1-197) and PP1 were present in both immunoprecipitates. A point mutation in the core PP1-binding sequence, KIKF, abrogated the association of GFP-NrbI-(F460A) with both PP1 and I-2. This established that even in cells, NrbI required PP1 binding to recruit I-2.
Co-localization of Neurabin II and I-2 in Cells-Prior immunocytochemical studies showed that NrbII was concentrated at adherens junctions in Madin-Darby canine kidney (MDCK) polarized epithelial cells (41). Immunocytochemistry using an anti-NrbII antibody confirmed that in MDCK II cells, NrbII was concentrated at the cell periphery consistent with adherens junctions (Fig. 8A). Double staining with anti-I-2 antibody showed a punctate distribution of I-2 throughout the cytoplasm, but a significant amount of I-2 was concentrated at the periphery of adherent cells (Fig. 8A) in regions that also stained for F-actin (data not shown). Merging of the images Following actin sedimentation and extraction as described under "Materials and Methods," the F-actin-bound proteins were analyzed by SDS-PAGE. Actin and NrbI-(1-516) were visualized by protein staining with Coomassie Blue, and the other proteins were immunoblotted with antibodies against PP1 and I-2. B shows identical F-actin sedimentation assays using 2 g of I-2-(1-114). C incorporates the data into a model that shows independent and mutually exclusive association of PP1 and I-2 with neurabins at the actin cytoskeleton. A conformation change in PP1 and/or I-2 converts this complex (indicated by arrow) into one where the association of one or both proteins is promoted or stabilized with the actin-bound neurabins.
established that NrbII and I-2 colocalized at the actin-rich adherens junctions in polarized epithelial cells.
NrbI and NrbII are highly expressed in neurons and concentrated in the post-synaptic density (34). Immunocytochemistry of cultured rat hippocampal neurons showed that I-2 was distributed throughout the cytoplasm and concentrated in puncta representing spines (Fig. 8B). I-2 staining overlapped with that for NrbII, which was even more concentrated in spines (16). Higher magnification (Fig. 8C) emphasized the localization of I-2 and NrbII in foci representing spines. I-2 and NrbII also colocalized at growth cones in younger neurons (data not shown). The data suggested that neurabins targeted I-2 to actin-rich structures in both neurons and non-neuronal cells. DISCUSSION Two classes of PP1 regulators, termed targeting subunits and inhibitors, control dephosphorylation events that regulate eukaryotic cell physiology (4,5). Previous data (4, 6) suggested a mutually exclusive interaction of these regulators with the PP1 catalytic subunit, based on the presence of a consensus motif, (K/R)(I/V)XF, required for PP1 binding and/or regulation. PP1 regulators also contain other sequences, such as ankyrin (42) and leucine-rich (43) repeats, which can contribute to PP1 binding. This suggests that by utilizing a subset of interactions with PP1, targeting subunits and inhibitors may not always compete but instead collaborate in PP1 regulation. CPI-17 (protein kinase C-potentiated inhibitor protein of 17 kDa) and the structurally related PHI (phosphatase holoenzyme inhibitor) lack recognizable KVXF motifs and inhibit myosin phosphatases composed of PP1 bound to a myosinbinding subunit (11,12). Several protein kinases regulate the smooth muscle myosin phosphatase activity by phosphorylating either CPI-17 (44) or myosin-binding subunit (45), providing direct evidence that these two PP1 regulators cooperate in the hormonal control of smooth muscle contraction.
Another example of coordinated control of PP1 activity by more than one regulator comes from the discovery that the growth arrest and DNA damage-inducible protein, GADD34, binds both PP1 and I-1, a PKA-activated PP1 inhibitor (10). Whereas GADD34 and I-1 both contain KIXF motifs essential for PP1 binding, independent interactions between the three proteins override potential competition for this key PP1-binding site and allow effective regulation of the GADD34-bound PP1 by the PKA-phosphorylated I-1 (10). In this study, we identified another heterotrimeric PP1 complex containing I-2 and neurabins. Unlike the GADD34⅐PP1⅐I-1 complex, PP1 is a necessary bridge in the assembly of the Nrb⅐PP1⅐I-2 complex. Several lines of evidence support this conclusion. First, GST-NrbI-(436 -479) consisting of 43 amino acids representing less than 5% of the full-length NrbI bound both PP1 and I-2, and mutating a single amino acid abolished NrbI binding to both proteins. Furthermore, no other proteins appeared to be required as the heterotrimeric complex could be reconstituted using purified PP1, I-2, and recombinant Nrb polypeptides.
Interestingly, deletion of the N-terminal actin-binding domain was required for NrbI-PP1 association in a yeast twohybrid assay. As the GAL4-NrbI was targeted to the yeast Neurabin⅐PP1⅐I-2 Complex Binds F-actin nucleus, 2 this suggested that the actin-binding domain inhibited NrbI association with PP1. However, biochemical studies showed that NrbI-(1-516) containing a functional actin-binding domain targeted both PP1 and I-2 to F-actin. Thus, F-actin may be a key factor in the assembly of a cellular NrbI⅐PP1⅐I-2 complex. I-2-(1-114) like WT hI-2 did not require PP1 but could associate directly with the actin-bound NrbI-(1-516). Indeed, I-2-(1-114) competed with PP1 for NrbI-(1-516) binding. However, the inclusion of the C-terminal sequences conserved in I-2 and I-4 containing a low affinity PP1-binding site (20) may stabilize the association of PP1 and I-2 with the actin-bound neurabins. In this regard, PP1 docked at NrbI was locked into place by I-2, and the "dock and lock" mechanism promoted PP1 recruitment to F-actin.
Prior studies (30) identified a docking site (colored red in Fig.  5B) on PP1 that accommodated an N-terminal sequence, IKGI, conserved in many I-2 isoforms. Mutations in the IKGI-binding site attenuated PP1 inhibition by I-2 (30). These mutant PP1 catalytic subunits bound NrbI-(103-1095) like WT PP1 in the yeast two-hybrid assay (data not shown). This suggested distinct recognition sites for NrbI and I-2 on the PP1 catalytic subunit. Thus, we analyzed a panel of 39 mutant yeast PP1 catalytic subunits and identified additional residues that impaired I-2 but not NrbI binding. Yet other residues (blue in Fig.  5B) abrogated the association of PP1 with NrbI-(103-1095) but not I-2. I-2 and NrbI also shared common docking sites on PP1 (green in Fig. 5B). These include the hydrophobic pocket that bound the tetrapeptide motifs (6), KIKF in NrbI (21,46) and KLHY in I-2 (20,47), and the ␤12-␤13 loop, which promoted I-2 (48) and NrbII (21,49) binding. These data suggested that a subset of these unique and overlapping sites are utilized for PP1 binding to neurabins and I-2 within a heterotrimeric complex.
MC, a cyclic heptapeptide inhibitor, docks in the PP1 catalytic site (6,48,50) and competes with I-2 for PP1 binding (20). MC abolished I-2 binding to an NrbI⅐PP1 complex formed by incubating GST-Nrbs with rat brain DOX lysates and enhanced PP1 recruitment by GST-Nrbs (Fig. 4). The ability of MC to displace I-2 from PP1 bound to the neurabin central domains argued that in this complex I-2 occupied the PP1 catalytic site and produced an inactive protein phosphatase. Yet, NrbI complexes immunoprecipitated from cultured mammalian cells were shown to be active protein phosphatases (14). This may indicate that the heterotrimeric complexes formed by full-length NrbI are different from those assembled by the truncated Nrb central domains or suggest that cellular mechanisms activate an NrbI⅐PP1⅐I-2 complex. One such mechanism may involve the phosphorylation of I-2 on threonine 72. In vitro studies showed that GSK-3 (51), mitogen-activated protein kinases (53), and CDK5, a neuronal protein kinase (18), phosphorylated I-2 within a dimeric PP1⅐I-2 complex and transiently activated this "ATP-Mg-dependent protein phosphatase" (51,52). Alternately, NrbI can be phosphorylated by PKA in vitro (55) and in vivo (14). This phosphorylation occurs at serine 461, immediately adjacent to the PP1-binding motif, and attenuates PP1 binding to NrbI (55). Thus, as shown in this study with the prototypic PP1-targeting subunit, G M , NrbI phosphorylation by PKA may also displace I-2. Finally, neurabins shuttle on and off the actin cytoskeleton in response to the Rac-1 GTPase by growth factors (54) providing yet another mechanism for regulating the Nrb⅐PP1⅐I-2 complex.
Immunocytochemistry established that I-2 and NrbII colocalized at actin-rich adherens junctions that mediate cell-cell contact in MDCK-polarized epithelial cells and in dendritic spines and growth cones of cultured rat hippocampal neurons.
Whereas the function of PP1 at the adherens junctions remains unknown, extensive studies suggest that PP1 dephosphorylates receptors, ion channels, and other signaling molecules in dendritic spines (14,16,56) to regulate synaptic transmission (57). Thus, the disruption of the mouse NrbII gene resulted in altered regulation of two prominent PP1 substrates, the ␣-amino-3-hydroxy-5-methyl-4-isoxazolepropionate and N-methyl-D-aspartate receptors, and altered synaptic plasticity (33), a cellular form of learning. This may suggest that the neurabin⅐PP1⅐I-2 complex controls the function and morphology of neurons and non-neuronal cells.
In conclusion, we identified a novel PP1 complex containing the two neurabin isoforms (NrbI or NrbII) and I-2. By using purified proteins, cultured cells, and tissue extracts from several species including human, mouse, rat, dog, and rabbit, we established that neurabins target the PP1⅐I-2 complex to Factin. The widespread expression of PP1, I-2, and NrbII suggests that this complex may regulate cytoskeletal functions and cell morphology in many tissues. Our data suggest that PP1 and I-2 are also recruited by other targeting subunits, including G M , the skeletal muscle glycogen targeting subunit. Thus, analysis of the Nrb⅐PP1⅐I-2 complex may provide important insights in the collaborations between I-2 and other PP1 regulators that dictate PP1 functions in mammalian tissues.